A Case Study of Ships Forming and Not Forming Tracks in Moderately Polluted Clouds

8/7/2019 A Case Study of Ships Forming and Not Forming Tracks in Moderately Polluted Clouds

1/19

15 AUGUST 2000 2729N O O N E E T A L .

2000 American Meteorological Society

A Case Study of Ships Forming and Not Forming Tracks in ModeratelyPolluted Clouds

KEVIN J. NOONE,a ELISABETH OSTROM,a RONALD J. FEREK,b,m TIM GARRETT,b PETER V. HOBBS,b

DOUG W. JOHNSON,c JONATHAN P. TAYLOR,c LYNN M. RUSSELL,d RICHARD C. FLAGAN,e JOHN H. SEINFELD,e

COLIN D. ODOWD,f MICHAEL H. SMITH,f PHILIP A. DURKEE,g KURT NIELSEN,g JAMES G. HUDSON,h

ROBERT A. POCKALNY,i LIEVE DE BOCK,j RENE E. VAN GRIEKEN,j

RICHARD F. GASPAROVIC,k AND IAN BROOKSl

a Department of Meteorology, Stockholm University, Stockholm, Swedenb Department of Atmospheric Sciences, University of Washington, Seattle, Washington

c Meteorological Research Flight, The Met. Office, Farnborough, Hampshire, United Kingdomd Department of Chemical Engineering, Princeton University, Princeton, New Jersey

e Department of Chemical Engineering, California Institute of Technology, Pasadena, Californiaf Centre for Marine and Atmospheric Sciences, University of Sunderland, Sunderland, United Kingdom

g Department of Meteorology, Naval Postgraduate School, Monterey, Californiah Desert Research Institute, University of Nevada, Reno, Nevada

i Graduate School of Oceanography, University of Rhode Island, Narragansett, Rhode Islandj Department of Chemistry, University of Antwerp, Wilrijk-Antwerp, Belgium

kApplied Physics Laboratory, The Johns Hopkins University, Laurel, Marylandl Scripps Institution of Oceanography, University of California, San Diego, La Jolla, California

(Manuscript received 11 July 1996, in final form 25 February 1998)

ABSTRACT

The effects of anthropogenic particulate emissions from ships on the radiative, microphysical, and chemical propertiesof moderately polluted marine stratiform clouds are examined. A case study of two ships in the same air mass ispresented where one of the vessels caused a discernible ship track while the other did not. In situ measurements ofcloud droplet size distributions, liquid water content, and cloud radiative properties, as well as aerosol size distributions(outside cloud, interstitial, and cloud droplet residual particles) and aerosol chemistry, are presented. These are relatedto measurements of cloud radiative properties. The differences between the aerosol in the two ship plumes are discussed;these indicate that combustion-derived particles in the size range of about 0.030.3-m radius were those that causedthe microphysical changes in the clouds that were responsible for the ship track.

The authors examine the processes behind ship track formation in a moderately polluted marine boundarylayer as an example of the effects that anthropogenic particulate pollution can have in the albedo of marinestratiform clouds.

1. Introduction

Clouds are one of the regulating factors controllingthe radiative balance of the earth. Marine stratiformclouds in particular play a significant role in the radi-ation balance due to their large horizontal extent, their

long lifetimes, and high reflectivity over an otherwiselow-reflectivity surface. The reflectivity of warm, non-precipitating marine stratiform clouds (their albedo) is

m Current affiliation: Office of Naval Research, Washington, D.C.

Corresponding author address: Dr. Kevin J. Noone, Department ofMeteorology, Stockholm University, S-106 91 Stockholm, Sweden.E-mail: [email protected]

a function of their optical depth. If the cloud has a rathernarrow droplet size distribution with a mean radiusmuch greater than the wavelength range of visible light,the optical depth of the cloud is a function of the cloudliquid water content, the density of the cloud droplets,the vertical thickness of the cloud, and the cloud drop

concentration (Twomey 1977).Droplet number concentration and liquid water con-

tent can vary independently of each other. Making agiven number of droplets larger by condensation of wa-ter vapor will increase the liquid water content withoutchanging droplet number. Such condensation requiresonly that droplets be present in a supersaturated envi-ronment. To increase droplet number, on the other hand,requires that additional particles be present upon whichnew cloud droplets can form or a supersaturation in-crease in the cloud.


2/19

2730 VOLUME 57J O U R N A L O F T H E A T M O S P H E R I C S C I E N C E S

a. Anthropogenic effects on cloud albedo

The link between pollution and planetary albedo waspointed out by Twomey (1974). The notion that an-thropogenically produced particles can have an influ-ence on cloud albedo has received a good deal of at-

tention in recent years. Twomey (1991) derived an ex-pression for the sensitivity of cloud reflectivity to chang-es in droplet number. He found that the most sensitiveclouds were those with low droplet number concentra-tions and a reflectivity of 0.5conditions typical ofmarine stratiform clouds.

The fact that the clouds for which the change in re-flectivity for a unit change in droplet number is largestare clouds with low droplet number concentrations hasmade anthropogenic influences on cloud albedo verydifficult to detect. An analysis by Schwartz (1988) couldnot detect a difference in mean albedo for Northern andSouthern Hemisphere clouds and thus could not detectany anthropogenic influence on cloud albedo. A sub-

sequent investigation (Kim and Cess 1993) of low-levelclouds off the east coasts of North America and Asiashowed enhanced cloud albedos in the coastal bound-aries where anthropogenic sulfate loadings are largest.

b. Ship tracks and susceptibility

Ship tracks have been observed in satellite imagesever since the early TIROS satellites became operationalthree decades ago (Conover 1966). Even in the earliestimages, the anomalous cloud lines were suspected tohave been caused by additional aerosol particles pro-duced by ships (Conover 1966; Twomey et al. 1968).

More recent observations have strengthened the linkbetween ships and anomalous cloud lines (Coakleyet al. 1987; Ferek et al. 1998; Hindman et al. 1994;King et al. 1993; Radke et al. 1989).

While the early observations of ship tracks in theTIROS images were limited to the visible region of thespectrum, later observations have been expanded to sev-eral wavelengths. The advanced very high resolutionradiometer (AVHRR) instrument on board the NationalOceanic and Atmospheric Administration (NOAA) po-lar orbiting satellites employs five different wave-lengths. While some ship tracks are apparent in the vis-ible wavelengths, many more become noticeable at 3.7m (channel 3 of the AVHRR). Platnick and Twomey

(1994) used AVHRR data to show that the sensitivityof cloud albedo to an increase in drop concentrationthe cloud susceptibilityof ship tracks was lower thanthe surrounding ambient clouds.

Ship tracks and anthropogenic perturbations of cloudalbedo are manifestations of the same underlying phe-nomenon. We can use ship tracks to learn about theprocesses behind changes in cloud albedo due to humanactivities. To this end, the Monterey Area Ship Track(MAST) experiment was designed. A specific aim ofthe project was to determine exactly what ship-induced

effect caused the ship track observable in satellite im-ages. The general aim was to investigate the processesbehind anthropogenic modification of cloud albedo.

c. Questions to be addressed

The MAST experiment was designed to test a set of10 hypotheses regarding the causes of ship track for-mation and how the tracks are maintained for long pe-riods in a convective boundary layer (Durkee et al.2000b). Tests of the various hypotheses are reported inother papers (e.g., Durkee et al. 2000c; Ferek et al. 2000;Durkee et al. 2000a). In this paper we will focus on acase study of two isolated pollution sourcestwo shipsencountered on 27 June 1994 in semipolluted condi-tions, one of which caused a measurable change in cloudproperties, while the other did not.

In order to understand the effects of the ship effluentson the clouds we sampled, we must also examine the

properties of the unperturbed clouds outside the areainfluenced by ship emissions. It is the differences be-tween the background clouds and the ship tracks thatwill provide information about the processes behindcloud albedo modification due to anthropogenic pollu-tion. We will begin by looking at the nature of the back-ground boundary layer in terms of its thermodynamicproperties, as well as the chemical and microphysicalproperties of the aerosol and cloud droplets in areasunaffected by ship emissions. We shall then compareand contrast the effects that two different ships in thesame air mass had on clouds in the marine boundarylayer with the aim of learning about the processes behindanthropogenic modification of cloud albedo.

The questions we will address here include the fol-lowing.

R What are the thermodynamic characteristics of themarine boundary layer in which the ship tracksformed?

R What is the chemical and microphysical nature of theaerosol in the background boundary layer?

R What is the chemical and microphysical nature of theaerosol particles that grew to form cloud droplets andof the interstitial aerosol in the background clouds?

R What differences were observed between the shiptracks and the background clouds in which theyformed?

R

What were the differences in ship emissions that couldhelp explain why one ship caused a ship track whilethe other did not?

R How can the results from this specific case study beused to improve our understanding of the more generalissue of anthropogenic influence on cloud albedo?

2. Experimental description

The MAST experiment took place in June 1994 offthe coast of central California. The experimental ap-


3/19


proach involved coordinated missions of several plat-forms aimed at testing a specific subset of the 10 hy-potheses mentioned above. The particular mission wasdefined each day depending on weather conditions, plat-form availability, and whether dedicated ships or shipsof opportunity were available.

A number of different platforms were involved in theMAST experiment: three aircraft [the University ofWashington (UW) Convair C-131A, Meteorological Re-search Flight (MRF) C-130 W Hercules, and the Na-tional Aeronautics and Space Administration ER-2], anairship (Naval Research Laboratory blimp), a researchvessel (the R/V Glorita), and NOAA and Defense Me-teorological Satellite Program satellites. In addition,various U.S. Navy vessels were available during certainperiods of the experiment for special maneuvers. Shipsof opportunity were investigated when navy vesselswere not available. A detailed description of the plat-forms is given elsewhere (Durkee et al. 2000b).

a. UW Convair C-131A

The primary aim of investigators on the C-131A wasto investigate the chemical and microphysical nature ofaerosols and clouds and aerosolcloud interactions. Theinstrumentation on the C-131A has been described indetail by Hobbs et al. (1991). We will present data pri-marily from four of the C-131 instruments: total particleconcentrations from a TSI Incorporated model 3760condensation particle counter, which counts particleslarger than 10-nm radius; accumulation-mode particleconcentrations from a wing-mounted Particle MeasuringSystems (PMS) passive cavity aerosol spectrometer

probe (PCASP-100X); droplet number concentrationfrom a PMS forward scattering spectrometer probe(FSSP-100X); and liquid water content from a GerberScientific Inc. Particulate Volume Monitor (PVM-100A). In addition to the C-131s normal payload, sev-eral guest investigators made measurements aboard theaircraft during MAST. Cloud condensation nuclei(CCN) measurements were made using a continuous-flow CCN spectrometer (Hudson 1989). Aerosol sizedistributions between 0.003- and 0.1-m radius weremeasured using a radial differential mobility analyzer[RCAD (Zhang et al. 1995)]. A modified version of theMICRON algorithm (Wolfenbarger and Seinfeld 1990;Russell et al. 1995) was used to invert the size distri-

bution measurements from the RCAD. The chemicaland microphysical properties of droplet residual aerosolparticles were measured using a counterflow virtual im-pactor (CVI) to sample the cloud drops (Noone et al.1988, Ogren et al. 1985).

b. MRF C-130 W Hercules (C-130)

The Meteorological Research Flight (MRF), a part ofthe U.K. Meteorological Office, operates a Royal AirForce C-130 Hercules aircraft that has been modified

to make it suited to a wide range of atmospheric re-search. A complete description of the standard meteo-rological instrumentation is given in Rogers et al.(1995), of the cloud physics instrumentation in Brown(1993) and Martin et al. (1994), and the radiation in-strumentation in Kilsby et al. (1992) and Saunders etal. (1992). The thermal volatility and size distributionsof aerosol particles were also measured aboard theC-130 (ODowd and Smith 1993).

The aim of the C-130 investigators was to investigatethe thermodynamic nature of the marine boundary layer,the microphysical and radiative properties of clouds, andthe microphysics and chemistry of aerosol particles.

c. R/V Glorita

The research vessel Glorita provided soundings ofthermodynamic parameters during the experiment. Inaddition, instruments aboard included a condensation

nuclei (CN) counter, a cloud radar system, a side-scan-ning lidar, and a ceilometer.

d. Ships investigated

We examined emissions from two different ships inthe same air mass: the USS Mount Vernon and the COS-CO Tai He. Some information about the two ships isgiven here; a more detailed description of the ships ispresented in Hobbs et al. (2000).

The USS Mount Vernon (LSD 39) is a landing shipwith a displacement of between 8600 and 13 700 tons.The Mount Vernon burns navy distillate fuel to powertwo 600-psi boilers in which superheated steam is gen-

erated. This steam in turn drives two geared steam tur-bines, each driving a separate shaft. The total power ofthe propulsion system is approximately 18 000 kW. TheMount Vernon was engaged in special maneuvers duringthe measurement period of 27 June. The ship had enteredthe operations area from the southeast, and at 0800 localtime (LT) began repeated legs alternately with andagainst the wind. The objective of this exercise was toexamine whether the wind speed relative to the shipaffected ship track formation. The ship steamed into thewind between 0800 and 1000 LT, with the wind between1000 and 1200 LT, against the wind again between 1200and 1500 LT, then with the wind between 1500 and 1700LT. She then steamed at a slight angle to the wind be-

tween 1700 and 1900 LT, in a north-northwesterly(NNW) direction, then south-southwesterly (SSW) be-tween 1900 and 2100 LT. She then exited the operationsarea in a northwesterly direction.

The Tai He is a container ship with a gross tonnageof 35 963 tons. She is powered by a six-cylinder, 17 000-kW diesel engine driving a single propeller. She has atop speed of 35 km h1 , a length of 236 m, a beam of32 m, and a draft of 12 m. During the measurementperiod, the Tai He was steaming in a southeasterly di-rection (120) at 32 km h1 .


4/19


FIG. 1. NOAA-9 3.7-m satellite image of the operations area at 1052 local time showing the easternPacific Ocean and California coast. The large trapezoid shows the operations area of the MRF C-130; thesmaller square represents the UW C-131A operations area. The Tai He course is indicated by the thick arrow.The R/V Glorita was on a northward leg during these operations, indicated by a line parallel to 125 W. TheTai He ship track is shown by a thin arrow.

3. Background marine boundary layercharacteristics

a. Thermodynamics and imagery

Figure 1 is an image from the NOAA-9 polar orbitingsatellite showing the operations area. The image showschannel 3 (3.7 m) of the AVHRR instrument on boardthe satellite. The coast of California appears in the upperright of the image. Monterey Bay is roughly in the centerof the coastline. An extensive deck of stratocumulusclouds is apparent over the Pacific Ocean, with someclearing near the coast. The operations areas of theC-130 and C-131A and the location of the R/V Glorita

are overlaid on the image. The approximate route of theTai He as she transited the operations area is also shown

on the figure.A number of ship tracks are apparent in the image,showing that several ships were causing tracks in thisair mass. The Tai He track in this image appears justwithin the rectangle showing the C-130 operations area,just to the left of the square denoting the area in whichthe C-131A flew (the Tai He track was also present invisible imagery). By the time the aircraft arrived onstation, the track had advected into the C-131A oper-ations area.

Knowing the ship position as a function of time and


5/19


FIG. 2. Vertical profiles of temperature and dewpoint temperature. The upper panel shows aprofile taken from the MRF C-130 during a descent to 30 m. Typical cloud-base and cloud-topheights are indicated by the shaded area in the upper panel. The lower panel shows two soundingstaken from the R/V Glorita, which was WSW of the aircraft operations area.

the mean wind speed and direction in the boundarylayer, a Gaussian puff dispersion model (Petersen andLavdas 1986) was used to calculate the location of theTai He plume as a function of time starting at 1100 localtime. The wind speed during the time the aircraft werein the vicinity of the Tai He varied between 10 and 20m s1 (mean value 14 m s1), and the wind directionvaried from NNW to NNE in the operations area. Thecalculated position of the Tai He plume at 1400 (towardthe end of the aircraft operations period) is shown inFig. 1 as a heavy line starting near the southeastern

corner of the C-131A operations area square. The cal-culated plume location corresponded extremely wellwith the ship track observed in the AVHRR channel-3satellite image from 1633 LT, corroborating that thetrack shown in Fig. 1 originated with the Tai He.

1) TEMPERATURE PROFILES

Figure 2 shows vertical profiles of temperature anddewpoint temperature for several locations and timesduring the operations period. The profiles show a strong


6/19


inversion at approximately 600 m. Some of the aircraftprofiles indicate that a slight decoupling of the boundarylayer may have taken place at around 250 m. There wasa layer of dry air above the cloud deck. The inversionheight increased as a function of distance from the coast.As is apparent in Fig. 1, the boundary layer was toppedby an extensive deck of stratocumulus clouds. The clouddeck was typically 200 m thick.

2) TURBULENCE

The C-130 flew a pattern upwind and to the north ofthe Tai He, aimed at obtaining turbulence data for theboundary layer. Heat and buoyancy fluxes were smalland slightly negative below 300 m. Both reached a max-imum of 3 W m2 at an altitude of 450 m and decreasedtoward cloud top. The Reynolds stress at the surfacewas 0.14 Pa and decreased with height. The MoninObukhov buoyancy parameter (z

i/L, where z

iis the

boundary layer depth and L is the Obukhov length) forthese conditions was 0.11, indicating that turbulence inthe boundary layer was shear driven from the surfacerather than buoyancy driven.

b. Aerosol properties

1) THERMAL VOLATILITY DATA

The thermal volatility of the 0.05 to 1-m-radius (ac-cumulation-mode) aerosol was characterized aboard theC-130, giving an indication of its chemical composition.The aerosol was sampled through an inlet on the C-130via a heating system to an optical particle counter (Par-

ticle Measuring Systems ASASP-X) where they werecounted and sized. This technique has been used in anumber of investigations (e.g., Clarke et al. 1987;ODowd and Smith 1993).

A typical run with this system was to first sample theambient aerosol and heat it to three temperatures (40,150, and 340C) in separate steps. Sulfuric acid vola-tilizes at approximately 100C, so changes in the aerosolsize distribution between 40 and 150C are interpretedas being due to loss of H 2SO 4 . Ammonium sulfate/bisulfate volatilizes between 150 and 340C, andchanges in the size distributions between these temper-atures are interpreted as being due to the loss of thesecompounds. The aerosol remaining at 340C is assumed

to be either sea salt or elemental carbon.There is a limitation to this technique when com-

pounds other than sulfuric acid, partially or totally neu-tralized sulfate, sea salt, or soot are present in the aero-sol. Organic compounds were found to have been pre-sent in the aerosol (Russell et al. 2000). The analyzedpolycyclic aromatic hydrocarbon components werefound in concentrations typically between 1 and 100 ngm3. These compounds typically have boiling points ator above 340C and could therefore be interpreted assea salt or elemental carbon with the volatility tech-

nique. There may also have been other lightermolec-ular weight organic compounds in the aerosol that vol-atilize at lower temperatures, thus potentially confound-ing the interpretation of the sulfuric acid/sulfate com-ponents. Since no additional information is available asto the presence or absence of such compounds in theaerosol, the following discussion will assume that theydo not contribute substantially to the thermal charac-teristics of the aerosol. We shall see in the Organiccloud droplet nuclei section [in section 3b(2) below],however, that carbon-containing particles played a sig-nificant role in cloud drop formation in the clouds weencountered.

Figure 3 illustrates the typical thermal volatility char-acteristics of the boundary layer and above-cloud aero-sol. Figure 3a shows size distributions at three temper-atures sampled at 30 m, well below cloud base in themarine boundary layer. The number size distribution ofthe aerosol between 0.05 and 1 m sampled at 40C

shows a peak at 0.08-m and a secondary peak at 0.2-m radius. The total accumulation-mode number con-centration was 138 cm3 for this distribution. The sizedistribution shifted to smaller sizes upon heating to150C, and the number concentration decreased to 100cm3. Upon heating to 340C, only 20-cm3 particleswere left, with the major decrease in number below 0.2-m radius.

The change between 40 and 150C is an indicationof the presence of some nonneutralized sulfuric acid,perhaps even an external mixture. The largest fractionof the accumulation-mode aerosol, however, was eitherpartially or totally neutralized sulfate. We assume herethat the particles remaining at 340C were sea salt. This

assumption is supported to a large extent by independentsingle particle analysis (next section.) Comparing the40 and 340C distributions, it is apparent that most ofthe particles larger than 0.2 m were sea salt; however,these particles only accounted for about 15% of thenumber of accumulation-mode aerosols. As is typicalfor the marine boundary layer, the number of accu-mulation-mode aerosol particles in this case was dom-inated by partially neutralized sulfate.

Figure 3b shows similar size distributions at 200 m,near cloud base. As would be expected for a well-mixedboundary layer, the size distributions and thermal vol-atility characteristics of the aerosol were the same asfor the sample at 30 m. Interstitial distributions mea-

sured in cloud at 400 m (not shown) indicated that bynumber approximately 80% of the accumulation-modeparticles had been taken up into cloud droplets.

Distributions for a sample taken above cloud in thefree troposphere are shown in Fig. 3c. There are markeddifferences in the composition and concentration of thefree tropospheric aerosol. On this day, the marineboundary layer was more polluted than the air aloft; theabove-cloud accumulation-mode concentration wasonly about 40 cm3 compared to about 120 cm3 in theboundary layera factor of 3 lower. There is also little


7/19


FIG. 3. Aerosol size distributions from the aerosol thermal volatilityinstrument. Size distributions for particles after heating to 40, 150,and 340C are shown in (a)(c). (a) and (b) show distributions in theboundary layer below cloud. (c) shows distributions taken in free-tropospheric air.

difference in the 40 and 150C distributions, indicatingthat the aerosol was mostly ammonium sulfate/bisulfateand probably well aged. There are particles left at 340C.In this case they are probably not dominated by sea salt(as discussed below in section 4.2.) Candidates for thisrefractory material other than sea salt include soil dustand elemental carbon or organic carbon compounds withhigh boiling points.

2) SINGLE PARTICLE ANALYSIS

Single particle analysis (SPA) was carried out on sam-ples of cloud droplet residual particles, particles in theTai He and Mount Vernon plumes, as well as the below-and above-cloud aerosol. Particle samples were placedin a scanning electron microscope (JEOL JSM-6300)and the size, morphology, and chemical composition ofthe particles was determined using electron probe X-raymicroanalysis techniques (Jambers et al. 1995). The

lower size limit for single particle analysis was 0.1-mradius; particles below this size were not resolved. Atotal of 500 particles per sample were analyzed. Onceparticle size and composition were determined, a hier-archical cluster analysis was performed that groupedindividual particles with similar compositions (Bernardet al. 1986).

Table 1 summarizes the results from the single particleanalysis. The first column in the table describes thesample. The subsequent columns give the number ofclusters obtained in the heirarchical cluster analysis, thepercentage of particles belonging to each cluster, themean radius of particles in the cluster, and the clustercomposition.

Two samples of residual particles from droplets in theunperturbed cloud were obtained during this flight. Onesample was obtained on the ferry flight out to the op-erations area, and the second was obtained in the op-erations area well away from the ships. Similar clusterswere found in both samples, which accounted for mostof the particles larger than 0.1 m: sea salt, particlescontaining organic material and chlorine, and particlescontaining sea salt and sulfur. The inorganic composi-tion of the particles is consistent with the compositionfor the boundary layer particles larger than 0.1 m de-termined using the thermal volatility technique. Twoother clusters were found in the sample on the ferryflight to the operations areaone containing only or-

ganic material, and the other containing sulfur, calcium,and chlorine. The additional two clusters in the ferryflight sample may have been a result of a slightly dif-ferent airmass composition near the coast as opposedto offshore near the ships.

Organic cloud droplet nuclei

The clusters containing organic material or organicmaterial plus chlorine are quite interesting. These com-positions account for between 25% and 36% of the cloud

droplet residual particles larger than 0.1 m in the twoambient cloud samples. The fact that organic materialalone was found in many of the particles indicates thatthe organic material was not associated with other aero-sol particles (i.e., as a coating). The organic plus chlo-rine cluster is also intriguing. This cluster may representa combination of organic material and sea salt. If thiswere the case, however, sodium should also be presentin the cluster, unless the amount of sea salt in the par-ticles was ver y small. Examination of individual spectra


8/19


9/19


TABLE 2. Summary of the bulk ionic chemical composition of theboundary layer aerosol (units are g m3). Two separate filter sampleswere analyzed by ion chromatography: one for total aerosol (wholeair inlet: particles droplets in cloud) and one for particles smallerthan 1.75-m radius (interstitial inlet). Samples were taken on boardthe C-131A and analyzed after the experiment at the University ofWashington.

Sample/Ion Cl Na nss-SO 4 NO3

NH4

Tai He track (interstitial)1.75 mTotal

0.060.04

0.040.03

0.040.01

0.060.05

0.010.02

Ambient (interstitial)1.75 mTotal

0.220.07

0.110.01

0.03

0.070.03

0.010.02

Tai He plume (below cloud)1.75 mTotal

1.41.59

0.88

0.50.65

0.270.16

0.50

Ambient (below cloud)1.75 mTotal

1.611.76

1.051.10

0.540.59

0.190.15

0.160.16

Mt. Vernon plume (below cloud)1.75 m

Total

1.49

1.57

0.89

1.01

0.76

0.66

0.17

0.19

0.32

0.34

TABLE 3. Summary of CCN measurements made in the backgroundmarine boundary layer and in ship plumes using the DRI CCN spec-trometer. CN are particles larger than 0.01-m radius; CCN are par-ticles active at 1% supersaturation; MDSc is the median supersatu-ration for the spectrum; K is the exponent in the equation N CSK

given for three different S ranges. M denotes a plume sample fromthe Mount Vernon and T denotes a sample from the Tai He plume.

Local timeCN

(cm3)CCN

(cm3 )MDSc

(%)

K

0.040.1

0.10.4

0.41.0

Background1204:271205:301327:001329:001341:241343:071351:401352:301412:001414:401426:011427:591431:161431:461435:311436:43

7501030150015601380111010201630

116283395484353443379491

0.390.300.300.300.240.300.240.30

1.11.21.21.11.21.01.11.1

0.560.400.530.450.420.360.410.41

0.290.230.300.240.240.240.230.27

PlumeT 1329:541330:07T 1338:171338:26

T 1351:191551:25M 1419:101419:20M 1421:071421:14M 1433:351433:42M 1434:541434:58M 1443:211443:30

11242149

2052671

2002812538607

0.390.39

0.510.390.510.390.390.39

1.31.4

1.01.21.31.21.11.2

0.680.74

0.860.500.610.580.500.55

0.250.39

0.390.340.510.300.330.32

sulfate), we obtain values between 1 and 20 g m3,depending on time and location in the boundary layer.Some of the variation in the mass estimated from thethermal volatility technique and CVI may have beendue to inlet effects; neither inlet was optimized to collectparticle mass.

4) CNN SPECTRA

Table 3 is a summary of the cloud condensation nuclei

measurements made using the Desert Research Institute(DRI) CCN spectrometer (Hudson 1989) in the back-ground marine boundary layer below cloud. The averageCCN/CN ratio for the eight samples was 0.3 (std dev 0.08)roughly one-third of the CCN in the boundarylayer were active as CCN at 1% supersaturation. Theaverage boundary layer CCN concentration of 368 cm3

(std dev 123) is a bit more than double the typicalcloud droplet concentration in the ambient clouds (about150 cm3see Table 5.) The CCN values reported hereare similar to previous observations made off the south-ern California coast (Hudson and Frisbie 1991). No CNdata for the ship plumes are given in Table 3 since theparticle counter used in the DRI CCN payload was not

able to measure the extremely high CN concentrationsin the plumes in the immediate vicinity of the ships.

The empirical power-law relation NCCN CSK is often

used to predict the number concentration of CCN [NCCN(cm3 )] active at a given supersaturation (S, percent),where C and K are constants that depend on air masstype. It has been observed that the exponent K is not aconstant over a large range of S; rather, it depends onthe range ofS values over which the slope is calculated.Values of K are given for three different S ranges.

The CCN concentrations in the ship plumes were all

well above background values (significance levels: TaiHe background, 99.99%; Mount Vernonback-ground, 98.7%; Tai HeMount Vernon: 95.06%). CCNconcentrations (at 1% supersaturation) in the Tai Heplume were a factor of 5 higher than in the backgroundboundary layer, while values in the Mount Vernon plume

were typically between 2 and 2.5 times greater, but attimes not much higher than background values. A de-tailed description of ship emissions is given in Hobbset al. (2000); in this particular case it is clear that whileboth ships were producing particles active at 1% su-persaturation, the Tai He CCN source strength was morethan double that of the Mount Vernon.

5) PARTICLE SCAVENGING

The fraction of aerosol material scavenged into cloudsis of interest from several points of view. From theaerosol perspective, only the fraction of the aerosolscavenged into cloud droplets will be subject to mod-

ification by aqueous-phase reactions in cloud water.Measurements have shown that the aerosol size spec-trum can be substantially modified by aqueous-phaseprocesses (Hoppel et al. 1994, 1986; Noone et al. 1992).Nucleation scavenging has been shown to be the pri-mary process for transferring aerosol material into clouddroplets (Flossmann et al. 1985; Noone et al. 1992).The nucleation step will thus control the initial chemicalcomposition of the cloud water, and to a large degreethe extent to which reactions can occur. Since the chem-ical composition and concentration of cloud water are


10/19


TABLE 4. Scavenging fraction for various aerosol components.Mass scavenging fractions were computed using ambient interstitialand ambient below-cloud samples from the UW C-131 interstitial(1.75-m radius) inlet (except nss-SO4

, for which the whole-airinlet data were used.) Here, N acc denotes accumulation-mode aerosolnumber and was calculated from UMIST size distributions obtainedon the MRF C-130.

Property Scavenged fraction

Cl massNa massnss-SO4

* massNO3

massNH4

massNacc

0.860.890.950.630.970.77

mass-dependent parameters, mass scavenging efficien-cies are of interest in this context. In the context of shiptrack formation, it will be less likely that an injectionof additional particles will lead to the production of asubstantial number of new cloud droplets if there isalready an overabundance of aerosol particles on whichcloud droplets can form in the boundary layer.

Table 4 shows the mass scavenging efficiency of thevarious major ionic components of the aerosol. Massscavenging efficiencies for sea salt were close to 90%,while nonsea salt sulfate was 95% scavenged intocloud droplets. The estimates were made from the datapresented in Table 2. In addition, the scavenging effi-ciency for accumulation-mode aerosol number is givenin Table 4. The value of 77% was calculated from ther-mal volatility measurements made at 400 m (just belowcloud base) and from cloud interstitial aerosol mea-surements. The observation that a large fraction of boththe accumulation-mode number and mass was taken upinto cloud droplets in the background clouds indicatesthat an additional source of particles in this size rangemay possibly have an effect on cloud microphysicalproperties.

4. Ship trackambient cloud comparisons

a. Tai He

1) TRANSECTS

Several transects were flown through the Tai He shiptrack. Summary data for five of the transects are pre-sented in Table 5. We shall discuss two transects in detail

to illustrate the typical changes in cloud and aerosolproperties between the ship track and the ambient cloud.

Three transects were flown with the UW C-131A fly-ing in cloud and the MRF C-130 flying above theC-131A. The objective was to have in situ measure-ments of cloud and aerosol properties using instrumentson the C-131A with simultaneous measurements of thecloud radiative properties using downward-looking ra-diometers on board the C-130 (two Eppley pyranome-ters measuring broadband hemispherical irradiances:one between 0.3 and 0.7 m, the other between 0.7 and

3.0 m). It should be noted that the two aircraft wereflying parallel to each other, but with a slight horizontaloffset during these transects. While in flight, shipplumes and tracks were identified by rapid increases anddecreases in total aerosol particle number concentra-tions. For the purposes of postflight analysis, track edgeswere determined by examination of the time series ofthe UW TSI 3760 particle counter. The midpoint of theparticle increase or decrease was identified, and a min-imum of 5 s of data on either side of the midpoint wasexcluded in order to assure that averages properly re-flected the track and out-of-track conditions.

There are a few facts to note when interpreting thetime series plots for the transects. The TSI particle coun-ter aboard the C-131A had a 5-s time lag with respectto the wing-mounted instruments (e.g., FSSP, PCASP,PVM-100). This lag has not been removed in the timeseries plots. Due to possible sampling artifacts due tointermittent droplet shattering using wing-mounted

PCASP measurements for these aircraft, we will focuson the relative concentration differences between theship track and the ambient cloud. We will not draw anyconclusions based on the absolute number from thePCASP measurements for these in-cloud transects.

We shall compare the differences between the shiptrack and ambient cloud in terms of several parameters:droplet number, cloud liquid water content, accumula-tion-mode aerosol concentration, total aerosol concen-tration, and droplet effective radius. These comparisonsare summarized in Table 5. The table includes the meanvalue, standard deviation, and number of observationsfor each of the parameters in the five transects. Effectiveradius values were calculated from the average drop size

distributions. The final column in the table gives theship track/ambient cloud ratio for each of the parame-ters. Mean values for the ship track were taken overperiods well away from the track edges to obtain a rep-resentative track average. Ambient cloud averageswere taken for equal periods (with one exception wherethe ambient cloud average was for a shorter period) wellaway from the track. In the following discussions, sta-tistically significant differences are all taken at the 95%confidence level.

(i) Transect-3 time series

The time series for transect 3 is shown in Fig. 4. The

Tai He plume was encountered at the end of the timeseries and is apparent as the broad peak in the particlemeasurements (Figs. 4d,e). Plume advection calcula-tions confirm that the peak in particle concentration be-tween roughly 1236 and 1237 LT was the Tai He plume,6070 min old and approximately 3.5 km wide. Clearincreases in total particle number, accumulation-modeparticle number, droplet concentration, and liquid watercontent in the ship track can be observed in this transect.A significant increase of 27% in droplet number wasobserved in this transect. Liquid water content also in-


11/19


TABLE

5.

Summaryo

fTaiHetransectaverages

.Averagevalueswereca

lculated

forequalorsim

ilar

intervals

inan

done

ithersi

deo

fash

iptrac

k.

Theun

itsare

Dropletnum

ber

(Nd

),tota

l

aeroso

l[N

a(T)]

,an

daccumu

latio

n-m

odeaeroso

l[N

a(P)]

cm3;

liqu

idwatercontent

[LWC(p)]

gm3;

drop

letef

fect

ivera

dius

(Reff

)m

.Rat

iosareta

kenasthetrackvalue

divided

by

theaverageo

fthetwoam

bientclou

dvaluesoneithersi

deo

fthetrac

k.

The

()co

lumng

ivesthevaluethat

,w

henaddedtoan

dsu

btracte

dfromthemean

,g

ivesthew

idtho

fthe

95%

con

fidence

interval

.

Transect

Parameter

Beforetrac

k

Mean

Std

dev

Count

Track

Mean

Std

dev

Count

Aftertrac

k

Mean

Std

dev

Count

Rat

io

1

Nd

LWC(p)

Na

(T)

Na

(P)

Reff

143 0

.291

828

187 7

.74

4 0.0

06

710 0

.09

12 0

.019

21

.9

31

.40

.29

28

28

28

28

28

1

49 0

.299

11

63

2

49 7

.75

6 0.0

07

38

15 0

.12

18

.80

.021

119

.7

47

.40

.38

28

28

28

28

28

140 0

.383

770

209 8

.39

4 0.0

13

413 0

.05

12

.10

.042

11

.3

39 0

.14

28

28

28

28

28

1.0

5

0.8

87

1.4

6

1.2

6

0.9

6

2

Nd

LWC(p)

Na

(T)

Na

(P)

Reff

n/a

n/a

n/a

n/a

n/a

n/a

n/a

n/a

n/a

n/a

n/a

n/a

n/a

n/a

n/a

n/a

n/a

n/a

n/a

n/a

1

75 0

.295

16

56

3

36 7

.22

7 0.0

07

68

25 0

.08

29

.20

.032

304

109

.50

.35

56

56

56

56

51

143 0

.262

929

168 7

.38

2 0.0

03

6 60.0

6

9.5

0.0

12

25

.6

25

.90

.25

56

56

56

56

51

1.2

2

1.1

3

1.7

8

2.0

0

0.9

8

3

Nd

LWC(p)

Na

(T)

Na

(P)

Reff

145 0

.226

858

114 7

.09

3 0.0

07

4 50.0

6

12

.40

.025

14

.1

17

.50

.24

42

42

42

42

42

1

82 0

.316

14

15

2

56 7

.24

7 0.0

10

55

21 0

.09

26

.80

.039

211

.6

82

.50

.34

42

42

42

42

42

141 0

.333

796

157 8

.06

6 0.0

18

7 50.0

4

20

.20

.063

22

.8

18

.10

.14

34

34

34

34

34

1.2

7

1.1

3

1.7

1

1.8

9

0.9

6

4

Nd

LWC(p)

Na

(T)

Na

(P)

Reff

151 0

.309

935

132 7

.75

3 0.0

07

4 50.0

6

10

.60

.03

14

.9

22

.50

.22

50

50

50

50

45

2

94 0

.287

24

36

3

29 6

.20

18 0

.009

156

34 0

.08

77

.70

.041

659

145 0

.29

50

50

50

50

45

143 0

.259

1000

124 7

.47

3 0.0

07

8 50.0

5

12

.50

.029

32

.1

21

.10

.18

50

50

50

50

45

2.0

0

1.0

1

2.5

2

2.5

7

0.8

1

5

Nd

LWC(p)

Na

(T)

Na

(P)

Reff

170 0

.194

990

71 7

.56

177 0

.017

12

16 0

.04

24 0

.024

17

22 0

.05

7 7 7 7 7

324 0

.259

65

32

8

30 6

.65

38 0

.009

577

180 0

.19

53 0

.012

806

252 0

.26

7 7 7 7 7

176 0

.194

972

94 7

.65

9 0.0

11

14

19 0

.04

12 0

.016

19

27 0

.06

7 7 7 7 7

1.8

7

1.3

4

6.6

6

10

.06

0.8

7


12/19


FIG. 4. Time series for transect 3 of the Tai He ship track. For these transects, the UW C-131Awas flying in cloud, while the MRF C-130 was flying a vertically stacked pattern above the clouddeck. Albedo measurements were made from the MRF C-130, while the data shown in (b)(e)

came from instruments on the UW C-131A. The star in (e) indicates the time at which an interstitialaerosol size distribution sample was taken.

creased by 13% but remained elevated even when par-ticle and droplet concentrations decreased at the edgeof the ship track. As in the first two transects, dropleteffective radius was larger in the ambient cloud afterthe ship track, but no trackambient cloud differenceswere observed.

An increase in cloud albedo of 14% was observed in

this case collocated with increases in particle and dropletconcentrations and liquid water content (LWC) in this tran-sect. The average visible albedo in the ambient clouds justbefore the ship track was 0.42 (std dev 0.023, n 299), while the average albedo in the ship track was 0.48(std dev 0.018, n 42). In the second transect weobserved an albedo increase that was broader than the


13/19


FIG. 5. Cloud and aerosol properties in transect 3 of the Tai Heship track. (a) shows cloud droplet size distributions for equal timeintervals before, in, and after the ship track. (b) shows cloud dropletresidual particle distributions for the corresponding periods.

observed microphysical changes in the cloud. The spatialscale of the albedo change in transect 3 is similar to theone observed in transect 2, while the spatial scales of theship track itself changed somewhat between transects.

Cloud and aerosol properties. Drop size distributionsbefore, in, and after the ship track are shown in Fig. 5a.As in the previous two transects, the distributions arequite similar. The variation in the ambient cloud distri-butions appears to be similar in magnitude to the var-iations between the ship track and ambient cloud. Re-sidual size distributions are shown in Fig. 5b. The re-sidual particle size distributions were measured using aPMS PCASP-X optical particle counter, which was cal-ibrated prior to the experiment (Ostrom et al. 2000). As

in the previous two transects, there was a slight increasein mode size for the residual particles in the ship track,though no marked changes in residual particle concen-trations were observed between the track and back-ground clouds. Interstitial aerosol size distributions werenot available for this transect.

(ii) Transect-4 time series

Figure 6 shows another transect of the Tai He trackat 1309 LT when the track was about 50 min old and

approximately 4 km wide. The differences in dropletand aerosol particle concentrations are more pronouncedhere than in the previous transects. Droplet number in-creased by a factor of 2 in the track, while total andaccumulation-mode particle concentrations were a fac-tor of 2.5 larger. Liquid water content showed a generaldecrease across the transect from 0.31 g m3 at the startto 0.26 g m3 at the end. In this case, there was a cleardecrease in droplet effective radius compared with theambient cloud on either side of the track. The effectiveradius decreased by 1.4 m (19%) in the track comparedwith the average for the ambient cloud samples. Nocorresponding albedo measurements are available forthis transect, since the C-130 and the C-131A were nolonger flying in formation at this time.

Cloud and aerosol properties. In contrast to the firstthree transects, there were marked changes in drop sizedistributions in this transect. Figure 7a shows averagedrop size distributions before, in, and after the track.

The track droplet size distribution shows a shift of themode toward smaller sizes shows a dramatic increasein the number of drops smaller than 7 m. This com-bination of decrease in drop size and increase in dropnumber is similar to previous observations in ship tracks(King et al. 1993; Radke et al. 1989). Figure 7b showsthe corresponding residual particle size distributions forthe transect. In this case, little change in the shape ofthe distribution is seen across the track. There is anincrease in residual particle number in the track but nomarked shift in size of the mode of the distribution.Interestingly, here there is a large change in the dropsize distribution and little change in the residual distri-bution, while in the previous transects little change was

observed in the drop distributions and an increase inmode size was seen for the residual particles. There aretwo possibilities for this relative lack of variation in theresidual distributions. One is that due to the shift tosmaller sizes of the droplet distribution, a larger fractionof the drops were below the cut size of the CVI, andthus the residual particles in these droplets would nothave been seen in the CVI. A second possibility is thatthe additional droplets in the track formed to a largeextent on particles smaller than 0.055 m. The latterscenario is supported by observations of the interstitialaerosol size distributions. Figure 7c shows interstitialsize distributions for the transect. While there is vari-ation in the background, the track distribution shows a

substantial increase in the number of particles between0.02 and 0.1 m. The greatest fractional increase forparticles in this range was at about 0.06 m. The TaiHe was thus emitting substantial numbers of particleslarge enough to have served as cloud droplet nuclei.

2) GENERAL DISCUSSION OF THE TAI HETRANSECTS

Five transects were flown through the Tai He track.Clear increases in total aerosol were observed in all


14/19


FIG. 6. Time series of measurements from the UW C-131A for transect 4 of the Tai He shiptrackas in Fig. 4. No albedo measurements were available for this transect.

transects, while similar increases in accumulation-modeaerosol and droplet number were observed in four ofthe five transects. In the first three transects, during allof which the plume was about 5060 min old, no sub-stantial changes were apparent in the droplet size dis-tributions. On the other hand, slight increases in residualparticle size were observed for these cases.

For the two transects closest to the ship, clear shiftstoward smaller sizes were observed in the droplet sizedistributions. Effective radii in track for these transectswere approximately 1 m smaller than in the surround-ing ambient clouds. These transects were quite similarto the ship tracks described in previous investigations(King et al. 1993; Radke et al. 1989). Interestingly, inthe cases where the largest changes in droplet distri-butions were observed, essentially no change was seenin the residual particle distributions.

Background interstitial aerosol distributions near the

Tai He were characterized by a broad peak betweenabout 0.01 and 0.05 m, and a second smaller and morevariable peak between 0.05 and 0.1 m. Interstitial aero-sol samples in the Tai He plumes showed large increasesin particles above 0.01 m, but primarily in sizes be-tween 0.02 and 0.1 m. CCN levels in the Tai He plumewere on average a factor of 5 greater than background

levels. Some of the track samples exhibited peaks in therange below 0.01 m as well.

b. Mount Vernon

Several transects through the Mount Vernon plumewere made by the UW C-131A shortly after 1500 LT.The Mount Vernon did not produce a track on any ofthe satellite images.


15/19


FIG. 7. Cloud and aerosol properties in transect 4 of the Tai Heship trackas in Fig. 5. Interstitial aerosol samples were taken before,in, and at the track edge in this transect.

1) TRANSECTS

Since all of the transects through the Mount Vernoneffluent in cloud were quite similar, only one (transect

2) will be discussed in detail here. Since the Mount

Vernon did not produce a track in the satellite imagery,

the term track is inappropriate. We shall use the term

plume in the following discussions when referring to

the aerosol plume encountered in cloud downwind of

the Mount Vernon.

Transect 2

Transect 2 of the Mount Vernon in-cloud plume wasflown between about 1517 and 1520 LT for a total of175 s.

Time series. Figure 8 shows the time series for tran-sect 2 of the Mount Vernon plume, approximately 2 kmwide. Again, several peaks were seen in the total particleconcentration trace shown in Fig. 8d. As in the other

transects, no corresponding changes were observed inthe other parameters. Accumulation-mode aerosol againshowed fair variability, with an average of 70 cm3 (stddev 32.2) across the transect. Droplet number was

again fairly constant at 194 cm3 (std dev 22.2), whileLWC was more variable around a mean of 0.179 g m3

(std dev 0.039).

Cloud and aerosol properties. Figure 9 summarizesthe microphysical changes through the Mount Vernonplume encountered between 1519:00 and 1519:24 LT

(shaded in Fig. 8d.) Figure 9a shows drop size distri-butions in and outside the plume. No change in dropsize distribution at all was observed in this pass. Thesame can be said of the residual particle size distribu-tions, as shown in Fig. 9b. Two interstitial aerosol size

distributions were taken on this pass (indicated as starsin Fig. 8d): one in the ambient cloud and one in the

Mount Vernon plume. The ambient cloud distributionis similar to distributions observed neat the Tai He.There is a broad peak between 0.01 and 0.06 m anda second smaller peak between 0.06 and 0.1 m. Onedifference between the background aerosol near the TaiHe is the presence of a distinct third mode at 0.008 m,which was not seen in the other observations. TheMount

Vernon plume aerosol shows an increase in particlesprimarily in the range between 0.01 and 0.03 m, but

also in the 0.050.1-m range. There was an increasein cloud droplet residual particles in this size range aswell in and after the plume. We have no interstitial

sample after the plume, but the correspondence betweenthe residual and interstitial particles in this case, and thefact that the residual distribution after the plume wasthe same as in the plume, seem to indicate that the

variation in particles of this size may have been due tovariation in the background aerosol and not due toplumebackground differences.

2) GENERAL DISCUSSION OF MOUNT VERNONTRANSECTS

Despite large increases in total aerosol concentration,no changes in cloud microphysics were observed in the

Mount Vernon plume, and no ship track was observed

in the satellite images that could be attributed to theMount Vernon. Droplet number was relatively constantat about 200 cm3 through all of the transects, and LWCwas more variable around a mean of 0.18 g m3.


16/19


FIG. 8. Time series for transect 2 of the Mount Vernon plume. The area between the lines in(d) denotes the plume from which in-plume cloud droplet and residual particle size distributions(shown in Fig. 9) were derived.

5. Presence/absence of a ship track: Comparisons

Two ships close to each other in the same air mass

affected the cloud deck above them in very differentways. The Tai He produced a distinct ship track in sev-eral satellite images from 27 June, while the MountVernon did not. In this section we contrast the two casesto see which factors exhibited the largest differencesbetween the two cases.

a. Aerosol properties

Perhaps the largest difference in any of the aerosolproperties between the Tai He and the Mount Vernon

cases was in the interstitial aerosol. In the Tai He case,

the ship produced an abundance of particles of all sizes

above 0.01 m; in particular, the concentration of par-

ticles in the 0.030.1-m and accumulation-mode rang-es was substantially augmented in the Tai He ship tracks

compared to background values. In the Mount Vernon

case, similar increases in total particle concentrations

were observed, but no corresponding increases in ac-

cumulation-mode particles were seen. Interstitial aerosol

distributions in the Mount Vernon plume indicated that

the increase in total particle number was primarily in

particles smaller than circa 0.03 m. Apparently these

particles were too small to have activated under the


17/19


FIG. 9. Cloud droplet (a), droplet residual aerosol (b), and inter-stitial aerosol size distributions (c) for transect 2 of the Mount Vernonplume.

prevailing supersaturations in the clouds near the Mount

Vernon, while the larger particles from the Tai He wereable to grow to form cloud droplets. This conjecture issupported by an analysis of emissions from a large num-ber of ships encountered during the MAST experiment(Hobbs et al. 2000) and by the observation that the TaiHe produced more than double the number of CCNactive at 1% supersaturation than did the Mount Vernon.

While it is not surprising that smaller particles areless active in cloud drop formation, the observed dif-ferences in particle characteristics can provide infor-mation relevant to the broader issue of indirect radiative

forcing of aerosols. The differences in aerosol charac-teristics for the two ships allows us to estimate the small-est size of particle able to influence the microphysicsand radiative properties of the clouds we encounteredon this particular day. Ostrom et al. (2000) concludedthat the number population of cloud droplets for theentire MAST experiment was controlled by particlessmaller than 0.1-m radius and that this upper limit ofparticle size controlling droplet number did not varysignificantly with boundary layer pollution level. Onthis particular day, we can estimate that the lower sizelimit for particles controlling droplet number was about0.020.04-m radius.

Present attempts to model indirect forcing on a globalscale use various empirical relationships between aero-sol sulfate mass and cloud albedo, even though the phys-ics behind albedo change in clouds is a number-basedrather than a mass-based phenomenon. If we can de-termine a size range of aerosol particles that dominates

cloud droplet number for low-level marine clouds, itmay be able to improve the parameterizations used inglobal models by providing a more logical link betweenaerosol mass and particle number.

An interesting difference in particle chemistry fromsingle particle analysis between the Tai He and MountVernon was the presence of particles containing siliconand organic material in the Tai He plume but not in the Mount Vernon plume. The Tai He burned diesel fuel,and silicon is used in the distillation cracking process,so traces of silicon are present in diesel fuel. The MountVernon was driven by a steam turbine, which burns navydistillate fuel. No substantial differences in bulk analysisof the major ionic species were observed in the tracks

or plumes of the ships, indicating that the particles pro-duced by the ships (and in the Tai He case responsiblefor the ship track effect) were small and did not con-tribute in any tangible way to the particle mass in theboundary layer.

b. Cloud properties

Clear decreases in droplet size were observed in twoof the Tai He transects, and in none of the Mount Vernonpasses. Droplet concentrations in the background cloudsnear the Tai He were slightly lower than near the MountVernon: 150 versus 200 cm3, respectively. It is notlikely that this difference alone could account for thefact that the Mount Vernon did not produce a track while

the Tai He did. Ship tracks were observed in even morepolluted clouds during the experiment (Noone et al.2000). Cloud droplet number was not any more variablein either location. The relative standard deviation ofdroplet number was 8% near the Tai He and 11% nearthe Mount Vernon.

6. Summary and conclusions

We have presented a case study of ship track for-mation in semipolluted clouds off the coast of central


18/19


California. We have compared and contrasted the effectsthat two different ships in the same air mass had onclouds in the marine boundary layer. One of the ships(the Tai He) produced a distinct ship track in satelliteimagery, while the other ship (the USS Mount Vernon)did not. We wished to answer six questions regardingthe nature of the backgound clouds in which ship trackswere forming, the differences in cloud and aerosol prop-erties between the background clouds and a specific shiptrack, and the reasons why the Tai He caused a shiptrack and the Mount Vernon did not when both shipswere in the same air mass.

Concerning the properties of the background clouds,we estimate that sea salt particles accounted for ap-proximately 15% of the cloud droplet nuclei in the am-bient clouds. Organic particles could account for ap-proximately 6%9% of the cloud droplet nuclei. Therest of the cloud droplet nuclei (about 76%) were es-timated to be partially neutralized sulfate particles.

Aerosol mass scavenging efficiencies in the backgroundclouds were typically between 80% and 95%, while thenumber scavenging efficiency for accumulation-modeparticles was approximately 80%.

From the point of view of ship track formation, nodifferences in sea salt concentrations that we could linkto ship track formation were seen between the ambientclouds and ship tracks. The high values of the accu-mulation-mode number and mass scavenging efficien-cies (about 85%95% and 77%, respectively) in thebackground clouds indicate that the microphysical prop-erties of the clouds could be susceptible to an additionalsource of particles in this size range.

Perhaps the most important difference between the

two ships in terms of track formation was in the sizeof the submicrometer aerosol produced by the ships.The Tai He, burning diesel fuel, produced substantialnumbers of particles in the size range between 0.01 and0.1 m, while the Mount Vernon, driven by a turbineengine, produced particles generally smaller than 0.03m. The fact that the larger particles (accumulation-mode particles and particles down to circa 0.030.04m) produced by the Tai He were able to form addi-tional cloud droplets in the tracks was supported by thesignificantly higher CCN flux in the Tai He plume com-pared with the Mount Vernon plume. The observed dif-ferences allow us to estimate that the lower limit ofparticle size that could affect cloud properties on thisday was between 0.02- and 0.04-m radius.

Acknowledgments. We are indebted to the pilots, cap-tain and crews of the UW C-131A, the MRF C-130, theNRL airship, and the R/V Glorita for their excellentwork during the field campaign. Drs. Wendell Nuss andChuck Wash at NPS provided valuable forecasting andmeteorological analysis during the campaign. This pro-ject was funded by the Office of Naval Research. Thesupport provided by Bob Bluth from ONR was indis-

pensable, and the project would not have been a successwithout his efforts.

REFERENCES

Bernard, P., R. Van Grieken, and D. Eisma, 1986: Classification ofestuarine particles using electron microprobe analysis and mul-tivariate techniques. Environ. Sci. Technol., 20, 467473.

Brown, P. R. A., 1993: Measurements of the ice water content incirrus using an evaporative technique. J. Atmos. Oceanic Tech-nol., 10, 579590.

Clarke, A. D., N. C. Ahlquist, and D. S. Covert, 1987: The Pacificmarine aerosol: Evidence for natural acid sulfates. J. Geophys.

Res., 92D, 41794190.Coakley, J. A., Jr., R. L. Bernstein, and P. A. Durkee, 1987: Effect

of ship-stack effluents on cloud reflectivity. Science, 237, 10201022.

Conover, J. H., 1966: Anomalous cloud lines. J. Atmos. Sci., 23, 778785.

Durkee, P. A., and Coauthors, 2000a: Composite ship track charac-teristics. J. Atmos. Sci., 57, 25422553., K. J. Noone, and R. T. Bluth, 2000b: The Monterey Area Ship

Track experiment. J. Atmos. Sci., 57, 25232541., and Coauthors, 2000c: The impact of ship-produced aerosolson the microstructure and albedo of warm marine stratocumulusclouds: A test of MAST hypotheses 1i and 1ii. J. Atmos. Sci.,57, 25542569.

Ferek, R. J., D. A. Hegg, and P. V. Hobbs, P. A. Durkee, and K.Nielsen, 1998: Measurements of ship-induced tracks in cloudsoff the Washington coast. J. Geophys. Res., 103, 23 19923 206., and Coauthors, 2000: Drizzle suppression in ship tracks. J.

Atmos. Sci., 57, 27072728.Flossmann, A. I., W. D. Hall, and H. R. Pruppacher, 1985: A theo-

retical study of the wet removal of atmospheric pollutants. PartI: The redistribution of aerosol particles captured through nu-cleation and impaction scavenging by growing cloud drops. J.

Atmos. Sci., 42, 583606.Hindman, E. E., W. M. Porch, J. G. Hudson, and P. A. Durkee, 1994:

Ship-produced cloud lines of 13 July 1991. Atmos. Environ., 28,

33933403.Hobbs, P. V., L. F. Radke, J. H. Lyons, R. J. Ferek, D. J. Coffman,

and T. J. Casadevall, 1991: Airborne measurements of particleand gas emissions from the 1990 volcanic eruptions of MountRedoubt. J. Geophys. Res., 96, 18 73518 752., and Coauthors, 2000: Emissions from ships with respect totheir effects on clouds. J. Atmos. Sci., 57, 25702590.

Hoppel, W. A., G. M. Frick, and R. E. Larson, 1986: Effect of non-precipitating clouds on the aerosol size distribution in the marineboundary layer. Geophys. Res. Lett., 13, 125128., , J. W. Fitzgerald, and B. J. Wattle, 1994: A cloud chamberstudy of the effect that nonprecipitating water clouds have onthe aerosol size distribution. Aerosol Sci. Technol., 20, 130.

Hudson, J. G., 1989: An instantaneous CCN spectrometer. J. Atmos.Oceanic Technol., 6, 10551065., and P. R. Frisbie, 1991: Cloud condensation nuclei near marinestratus. J. Geophys. Res., 96, 20 79520 808.

Jambers, W., L. De Bock, and R. Van Grieken, 1995: Recent advancesin the analysis of individual environmental particles. Analyst,120, 681692.

Kilsby, C. G., D. P. Edwards, R. W. Saunders, and J. S. Foot, 1992:Water vapour continuum absorption in the tropics: Aircraft mea-surements and model comparisons. Quart. J. Roy. Meteor. Soc.,118, 715748.

Kim, Y., and R. D. Cess, 1993: Effect of anthropogenic sulfate aero-sols on low-level cloud albedo over oceans. J. Geophys. Res.,98, 14 88314 885.

King, M. D., L. F. Radke, and P. V. Hobbs, 1993: Optical propertiesof marine stratocumulus clouds modified by ships. J. Geophys.

Res., 98, 27292739.


19/19


Martin, G. M., D. W. Johnson, and A. Spice, 1994: The measurementand parameterization of effective radius of droplets in warmstratocumulus clouds. J. Atmos. Sci., 51, 18231842.

Noone, K. J., J. A. Ogren, J. Heintzenberg, R. J. Charlson, and D.S. Covert, 1988: Design and calibration of a counterflow virtualimpactor for sampling of atmospheric fog and cloud droplets.

Aerosol Sci. Technol., 8, 235244.

, and Coauthors, 1992: Changes in aerosol size and phase dis-tributions due to physical and chemical processes in fog. Tellus,44B, 489504., and Coauthors, 2000: A case study of ship track formation ina polluted marine boundary layer. J. Atmos. Sci., 57, 27482764.

ODowd, C. D., and M. H. Smith, 1993: Physicochemical propertiesof aerosols over the northeast Atlantic: Evidence for wind speedrelated submicron sea-salt aerosol production. J. Geophys. Res.,98, 11371149.

Ogren, J. A., J. Heintzenberg, and R. J. Charlson, 1985: In-situ sam-pling of clouds with a droplet to aerosol converter. Geophys.

Res. Lett., 12, 121124.Ostrom, E., K. J. Noone, and R. A. Pockalny, 2000: Cloud droplet

residual particle microphysics in marine stratocumulus cloudsobserved during the Monterey Area Ship Track experiment. J.

Atmos. Sci., 57, 26712683.Petersen, W. B., and L. G. Lavdas, 1986: Inpuff 2.0A multiple

source Gaussian puff dispersion algorithm. EPA/600/8-86/024,U.S. Environmental Protection Agency, Research Triangle Park,NC, 110 pp.

Platnick, S., and S. Twomey, 1994: Determining the susceptibility ofcloud albedo to changes in droplet concentration with the Ad-vanced Very High Resolution Radiometer. J. Appl. Meteor., 33,334347.

Radke, L. F., J. A. Coakley Jr., and M. D. King, 1989: Direct andremote sensing observations of the effects of ships on clouds.Science, 246, 11461149.

Rogers, D. P., D. W. Johnson, and C. A. Friehe, 1995: The stableinternal boundary layer over a coastal sea. Part I: Airborne mea-surements of the mean and turbulence structure. J. Atmos. Sci.,52, 667683.

Russell, L. M., R. C. Flagan, and J. H. Seinfeld, 1995: Asymmetricinstrument response resulting from mixing effects in acceleratedDMA-CPC measurements. Aerosol Sci. Technol., 23, 491509., K. J. Noone, R. J. Ferek, R. A. Pockalny, R. C. Flagan, andJ. H. Seinfeld, 2000: Combustion organic aerosol as cloud con-densation nuclei in ship tracks. J. Atmos. Sci., 57, 25912606.

Saunders, R. W., G. Brogniez, J. C. Buriez, R. Meerkotter, and P.Wendling, 1992: A comparison of measured and modeled broad-band fluxes from aircraft data during the ICE 89 field experi-ment. J. Atmos. Oceanic Technol., 9, 391406.

Schwartz, S. E., 1988: Are global cloud albedo and climate controlledby marine phytoplankton? Nature, 336, 441445.

Twomey, S., 1974: Pollution and the planetary albedo. Atmos. En-viron., 8, 12511256., 1977: Atmospheric Aerosols. Elsevier Scientific, 302 pp., 1991: Aerosols, clouds and radiation. Atmos. Environ., 25A,24352442.

, H. B. Howell, and T. A. Wojciechowski, 1968: Comments onAnomalous cloud lines. J. Atmos. Sci., 25, 333334.

Wolfenbarger, J. K., and J. H. Seinfeld, 1990: Inversion of aerosolsize distribution data. J. Aerosol Sci., 21, 227247.

Zhang, S.-H., Y. Akutsu, L. M. Russell, R. C. Flagan, and J. H.Seinfeld, 1995: Radial differential mobility analyzer. AerosolSci. Technol., 23, 357372.

A Case Study of Ships Forming and Not Forming Tracks in Moderately Polluted Clouds

Documents

Transcript of A Case Study of Ships Forming and Not Forming Tracks in Moderately Polluted Clouds